VTKDataIO.jl

A package for reading, writing and visualizing mesh data, and interfacing with Python's VTK.
Author JuliaTopOpt
Popularity
23 Stars
Updated Last
2 Years Ago
Started In
July 2017

VTKDataIO

Overview

VTKDataIO.jl presents a number of input, output and visualization functionalities for geometric meshes with point and cell scalar and vector data. This module attempts to bridge between VTKDataTypes.jl, Julia's native module for representing VTK data types and the Visualization Toolkit (VTK) hence giving access to many of VTK's capabilities to Julia users.

Python version

VTKDataIO.jl was only tested with Python 3.8. Make sure PyCall is buit to use Python 3.8.

Summary of capabilities

Input and output

You can use VTKDataIO.jl to read/write any of the following file formats into/from the corresponding type in VTKDataTypes.jl: vtk, vtu, vtp, vts, vtr, vti, vtm, pvd, stl, ply. You can do this using read_vtk, write_vtk, read_stl, write_stl, read_ply, and write_ply.

Visualization and rendered output

You can use VTKDataIO.jl to visualize a scalar or vector field with a 3D heat map and a legend. This includes point-based coloring or cell based coloring, as well as other features such as wireframe and glyph representations. The resulting visualization can be written to ply or x3d formats. You can do this using visualize, write_x3d, and write_ply. visualize_3ds can also be used to visualize 3ds files directly.

Interfacing with Python's VTK

You can use PyVTK to change a Julia native VTK object to a VTK data PyObject that can be used in your own VTK pipeline through PyCall. You can also use _VTKDataTypes to change a VTK data PyObject back to Julia's native VTK types.

Test cases

If you run the following code, you should be able to get a cubic mesh with randomly coloured cells.

using VTKDataTypes
using VTKDataIO

x = y = z = [-2, -1, 0, 1, 2];
rect = VTKRectilinearData((x, y, z));
rect.cell_data["Cell scalar"] = reshape([rand() for i in 1:num_of_cells(rect)], cell_extents(rect));
visualize(rect, color="Cell scalar")

image

You can also do point-based coloring as so:

rect.point_data["Point scalar"] = reshape([rand() for i in 1:num_of_points(rect)], extents(rect));
visualize(rect, color="Point scalar")

image

Point based vector data can be represented by arrows using the representation option.

rect.point_data["Point vector"] = reshape([2*rand()-1 for i in 1:3*num_of_points(rect)], (3, extents(rect)...));
visualize(rect, color="Point vector", representation=:glyph, scale_factor=0.5)

image

Wireframe representation can be used as follows:

visualize(rect, color="Point scalar", representation=:wireframe)

image

If you have the RGB colors data directly, you can inform the visualize function to use it using the RGB = true option as so:

rect.point_data["RGB colors"] = reshape([round(255*rand()) for i in 1:3*num_of_points(rect)], (3, extents(rect)...));
visualize(rect, color="RGB colors", RGB=true)

image

Used By Packages

No packages found.